BCAL Showers in DANA

I have recently taken a first, consumers-eye look at the BCAL shower reconstruction using the GlueX DANA framework. The plots shown below only look at the output of the DBCALShower factory obtained a'la:

vector showers;
loop->Get(showers);

Comparisons are made to the "thrown" values of the initial particles as obtained from the DMCThrown factory.

Click on any of the plots below for a larger view.

This is a plot was made using the built-in generator in hdgeant that uses the KINE card in the control.in file:

KINE 1 1.0 30. 85. 0. 0. 65.

This threw a single 1 GeV photon for every event randomly in all directions from the center of the target (0,0,65).

The plot is of the ratio of thrown energy (always 1GeV in this case) to reconstructed energy as obtained from DBCALShower::Ecorr. The ratio is plotted as a function of distance along the beam axis(Z). The distance is from DBCALShower::z, the reconstructed z coordinate, but shifted by about 214cm to bring it in alignment with the hdgeant coordinate system.

The blue line is a parabola + offset that I simply eye-balled to the data. It is used to correct the value of Ecorr in the plots below.

This plot is the same as the one above, except that the correction function was applied to the reconstructed shower energy. This is now centered close to 1 for all values of Z as one would like.

Using ROOTs FitSlicesY(), a gaussian was fit to the previous plot for each x-bin projected onto the y-axis. This plot shows the sigma values for each of those plots.

The units of the values are fractional energy resolution. So, the ends have the worst resolution at around 8% or 80MeV in this case of 1GeV photons. The minimum is just over 3% or 30MeV here.

The design resolution of the detector is:

&sigma(E)/E = 5.4%/&radic E(GeV) ⊕ 0.7%

so for 1GeV photons, this evaluates to 5.5% or about 55MeV.

Note that here, the distribution of photon angles is opposite from what one might expect from beam data. Specifically, we expect beam data to be forward biased so our largest statistical sample will be in the forward direction. In this test, due to the geometry and isotropic photon distribution, the forward part of the BCAL has the least statitics.

This plot is of the invariant mass of 2 shower events in the BCAL. For this plot, the KINE card was changed to throw &pios instead of photons. Note: the &pios were themselves 1GeV so the resulting photons were each considerably less than 1GeV.

The lower, black distribution is the 2-&gamma invariant mass made without correcting the shower energies with the parabola function discussed above. The blue distribution with the peak is with the parabolic correction applied.

The fit is a gaussian that indicates an invariant mass resolution (grain of salt) of about 8MeV. The mean though is systematically off at 117MeV rather than 139MeV. This is expected to improve with a better parameterization of the correction based on z and with a possible correction based on reconstructed E.

Summary

  • The BCAL reconstruction code in DANA is able to reconstruct showers with both position and energy information, but some corrections to both are still required.
  • The energy resolution of the BCAL seems to be in the right ballpark for the design specs. Are there other factors not currently in the simulation though (electronics smearing, noise hits, etc...)?
David Lawrence
davidl@jlab.org
Last Updated October 23 2006 10:40:36 AM